Nitrogen oxide storage and regeneration on platinum/barium oxide/gamma-aluminum trioxide lean nitrogen oxide traps

Saurabh Sudhir Chaugule, Purdue University

Abstract

Lean NOx Traps (LNT), a leading automotive NOx abatement technology, have been under intense investigation from industry and academia for the past two decades. This work presents the results of our reaction experiments which simulate the cyclic exhaust conditions over monolithic Lean NOx Traps (LNT) samples, providing overall performance data, and in situ Diffuse Reflectance Infra-red Spectroscopy (DRIFTS) studies which provide information about the surface chemistry. Insights from these data, when combined, have enabled conception of physico-chemical models of the NOx storage-reduction processes upon which our mathematical models of trap performance are based. A set of seven LNT formulations with 0.6–6.3 wt.% Pt and 4–20 wt.% Ba on Al2O3 were investigated for their NO x trapping abilities at 300°C, 30000 hr−1. It was found that the initial complete NOx storage ability of the traps depends on the proximity of Ba to Pt and thus on the Pt, Ba loading and dispersion. It was also concluded that the NOx storage occurs via two parallel pathways—(1) using the spilled over oxygen as the oxidant for nitrate formation on Ba vicinal to Pt (fast route) and (2) using NO 2 itself as the oxidant via the disproportionation reaction on the remaining Ba sites (slow route). The NO2 disproportionation rate was also found to be susceptible to the NO2 partial pressure over the trap, rendering NOx storage less efficient when NO + O2 was used as the NOx source vs. NO2 + O2. DRIFTS showed that bulk-like nitrates were formed on samples with 20 wt.% Ba but existed primarily in the form of surface nitrates on 8 and 4 wt.% Ba samples. Furthermore, the presence of CO2 and H2O in the lean-rich feed led to competitive adsorption on Ba and led to formation of carbonates (20 wt.% Ba)/carboxylates (8 and 4 wt.% Ba), hydroxides and nitrates. In addition, presence of water was found to cause agglomeration of the highly dispersed Ba phase on 8 and 4 wt.% Ba. A phenomenological model was developed to explain all these observed effects on the variety of LNTs. The insights gained from the NOx storage experiments allowed us to speculate that having higher Pt loading near the inlet of the monoliths and lower loadings downstream might prove to be a formulation with higher trapping efficiency and lower cost as compared to the traps with constant Pt loadings across the monolith channel length. Regeneration was studied between 150–300°C with H2 as a model reductant. At 30000 hr−1, the regeneration was found to be H-atom feed rate limited. It was observed that NH3 was as good a reductant as H2 and was found that ammonia actually played the role of an intermediate during regeneration with H2. H2O or CO2 did not have any significant impact on the regeneration mechanism. A ‘plug flow’ (moving front) model for regeneration was proposed and based on this mechanism it was concluded that to achieve maximum selectivity towards N2, an incomplete regeneration of the trap was needed since NH3 escaped only in the absence of NOx downstream of the reductant front. In summary, the fundamental knowledge gained from the experiments could be extended to formulation of more efficient and cheaper LNT systems and optimum strategies for their operation in addition to being used as the building block for a mathematical model of the process. (Abstract shortened by UMI.)

Degree

Ph.D.

Advisors

Delgass, Purdue University.

Subject Area

Chemical engineering

COinS